Artificial intelligence and story-telling support: Do algorithms do It Better?

Abstract

Journalism has always been shaped by technology (Pavlik, 2000); however, the changes brought about by increasing automation and algorithms are having a profound impact on how news is produced and consumed (Thurman, Lewis & Kunert, 2019). More specifically, automation techniques and algorithmic technology are (re)shaping content production by means of automated storytelling, data mining, news dissemination and content optimization (Diakopoulos, 2019). This study sets out to illuminate how algorithms can enhance story telling capabilities for the production of feature stories. A common thread of criticism associated with online stories is their offering fragmented bits of information (Eveland, 2003) and reproducing news stories, the so-called phenomenon of churnalism (Saridou, Spyridou & Veglis, 2017). Recent work (Andersen & Strömbäck, 2021) concludes no general learning effects from online outlets (as opposed to offline media), a finding raising serious questions regarding a broadly informed citizenry in the web 3.0 era. A key question thus is how, under severe time pressures imposed by the new media ecosystem, should professional agency and algorithms be blended together in order to efficiently and effectively produce news stories which contain a diversity of sources and views and avoid repetition and banal positioning. The study draws data from the collaboration of JECT.AI and SigmaLive. JECT.AI is a company which has produced a tool enabling journalists to discover a multitude of relevant sources and data, and thus positions, during content creation on a real-time basis. Robust work on professional practices and norms argues that journalists tend to sustain and reproduce dominant practices as a means to delineate their professional status and work norms (Singer, 2015). These include among others, the use of familiar and well-established sources to create news stories; however, such practices tend to reduce the diversity of sources and minimize creativity and plurality, and eventually the angles used to communicate information. As a response to this trend, this tool aims to support journalists in automatically retrieving news information with creative strategies that codified the expertise of experienced journalists (Maiden et al, 2019). SigmaLive is the leading mainstream news player in the online news landscape of Cyprus. JECT.AI has adjusted its tool in the Greek language used by SigmaLive. After implementing the tool for a period of 60 days, two questionnaires were developed: one for the journalists aiming to assess their experience with the tool and identify the perceived benefits and drawbacks, and one for the users aiming to explore the level of reader satisfaction when consuming stories developed by using the tool. Findings are deemed important in two ways; first, by identifying which design parameters and algorithmic input ensure maximum efficiency gains in terms of speeding up the monitoring of information and expanding real-time access to a variety of sources and data. Second, by assessing how this tool shapes story production in practice; is content produced by the aid of algorithmic data mining perceived as ‘better’ when compared to content produced solely by professionals? Can algorithmic data mining result in higher quality and more unique journalism that can offer a competitive edge in the marketplace while catering for a well-informed citizenry?